We live in a three-dimensional (3D) environment, and accurate perception of the 3D structure of the visual scene is crucial for many daily activities. Although much is known about human perception of depth from multiple cues (such as binocular disparity, motion parallax, and texture), our understanding of the neural mechanisms of depth perception remains very incomplete. This proposal addresses three fundamental issues regarding the neural basis of depth perception. Aim #1 examines the neural representation of 3D surface structure based on smooth spatial variations in binocular disparity. We will characterize how neurons in extrastriate visual cortex code the orientation and spatial frequency of depth corrugations defined by disparity gradients, and we will explore whether linear or nonlinear receptive field mechanisms are involved in such selectivity. In addition, we will reversibly inactivate these brain regions to probe for causal links between neural activity and perception of disparity-defined surface structure. Aim #2 provides the first direct tests of a potential neural substrate for depth perception from motion parallax. We will train monkeys to discriminate depth from motion parallax, and will use single-unit recordings and electrical microstimulation to test the hypothesis that area MT plays an important functional role in this form of depth perception. Aim #3 tackles a major unexplored question: how do we detect objects moving in 3D space during self-motion? We have devised a novel behavioral task to demonstrate that detecting inconsistencies between two depth cues--disparity and motion parallax--provides a robust mechanism for detecting object motion, and we test the hypothesis that neurons in area MT with incongruent depth tuning for disparity and motion parallax play an important role in this process. This research addresses the general problem of how neural circuits extract specialized information from the visual scene that is computationally important for solving specific behavioral tasks, and thus has broad application to many problems in systems neuroscience. The proposed research is directly relevant to the research priorities of the Strabismus, Amplyopia, and Visual Processing program at the National Eye Institute.